Embedded Files
My last task revolved around some data that our supervisor collected last year in September. He had the opportunity to witness basking shark breaching and took that chance to write down the timestamp of each breach and the approximate area in which each of them happened. He wanted me to analyse the data and answer the question: "Is a breach more likely to happen when another breach has just happened?".
I added a video from National geographic on the right that quickly introduces the species and shows their astonishing breaches (1:25).
This assignment was quite the challenge and I felt like quitting often, but in the end I think I found a solution that can answer the question. In case the occurrence of a breach was independent of another breach the distribution of the time between them would be exponential, so the only thing that was left to do was look at our inter-breach time distribution and how well that confirms to a exponential distribution.
On the left you can see a histogram of the inter-breach length in function of its frequency in the dataset. The values ranged from 0 (the same timestamp) to 34 minutes.
We can instinctively see that the number of times the inter-breach period was less than 1 minute is much more (41.6%) than what would be expected from an exponential distribution. This was also confirmed with a Lilliefors test.
Scientific note
Scientific note
Because so little is known about these extraordinary animals and especially their breaching behaviour, every new observation can give new insights. This is why Simon has given me the opportunity to write a scientific note on the results I obtained with his data. We are still in the writing-phase, with him looking at my first draft right now. It would be absolutely amazing if I could come out of this master with a publication on my name, so I am eager to continue working on this opportunity.
Report abuse